| Abstrakt: |
Significant increases in automation and vehicle autonomy are expected to transform today’s operatorsof single unmanned vehicles performing single missions into tomorrow’s supervisorsof multiple autonomous systems performing multiple missions. What displays and tools will those future supervisors require, given that they will manage automation that is deterministic and brittle in the dynamic and unpredictable battlespaces of the future? To help mitigate the shortfalls of automation for the future supervisor, we focus on how to make the scopeof the events handled by automation explicit to the supervisor. Here, we explore the highlighting of different event attributes, including scope, to guide attention during supervision, and we investigate aspects of geospatial integration and tradeoffs in how the highlighting is applied. These concepts are developed and applied in the domain of unmanned aerial vehicles (UAVs) guided by path planning automation-mediated routes. In a study, participants monitored multiple concurrent missions, and triaged events varying in severity and automation scope, to decide when to intervene. Four different methods of guiding attention were contrasted. Events were (1) in a callout separatedfrom a map, (2) integratedinto the map, (3) map-integrated and categorizedby priority but not task-tailorable, or (4) map-integrated, categorized, and task-tailorable. Monitoring was fastest and most accurate for categorized and tailorable events, validating the utility of the attentional guidance. These two conditions traded off in the support they provided, however. Monitoring was significantly faster for tailorable vs. categorized events; however, a significant time cost for correctly configuring the tailorable guidance reduced its advantage. This study validates a novel approach for conveying automation scope, and elucidates the need to design and validate new concepts for the specific task needs of the future autonomous systems supervisor. |
